Categorized in: 

Utilize Deep Research in ChatGPT

Deep Research is a mode built into ChatGPT that does major research for you: it browses the web, reads through multiple sources and hands back a structured, cited report instead of a quick one-off answer.

What is Deep Research, and when should I use it?

Deep Research is best saved for questions that genuinely need more than a fast answer; think literature reviews, policy summaries or comparing a handful of tools or vendors side by side. For a quick fact or a simple question, regular ChatGPT is faster and works well.

CU System employees can run 25 deep research reports every 30 days.

How do I run a Deep Research report?

  1. Open a new chat in ChatGPT and select the + icon next to the message box.
  2. Choose Deep research from the Tools menu.

The start chat menu in ChatGPT. The Deep research option is highlighted

  1. Describe what you need. Be specific about your goal, scope and the format you want; a good prompt reads more like a short brief than a one-line question. For example: "Compare the pricing, features and security practices of three project management tools for a mid-sized university department, and summarize the trade-offs."
  2. ChatGPT will propose a research plan before it starts running. Take a moment to review it and adjust the scope if it's heading in the wrong direction.

Screenshot of ChatGPT completing a deep research prompt.

  1. Once you approve the plan, Deep Research gets to work. Reports typically take 5–30 minutes, so feel free to step away and come back. You'll get a notification when it's ready.

Screenshot of ChatGPT completing a deep research prompt. The progress bar is highlighted

  1. Review your finished report. It will include inline citations and source links, so it is easy to check anything before you use it in your own work.

A screenshot of a completed deep research response from ChatGPT

 

Remember: Deep Research is a major time-saver, but it isn't infallible. Always double-check important facts, figures or quotes against the original sources before relying on them.

Add new comment